RuntimeError: storage has wrong size

I got this in the prediction section (bmt) root@ABC001:/data/zuyuehan/project/BMT# python ./sample/single_video_prediction.py --prop_generator_model_path ./sample/best_prop_model.pt --pretrained_cap_model_path ./sample/best_cap_model.pt --vggish_features_path ./test/pandemic_vggish.npy --rgb_features_path ./test/pandemic_rgb.npy --flow_features_path ./test/pandemic_flow.npy --duration_in_secs 99 --device_id 0 --max_prop_per_vid 100 --nms_tiou_thresh 0.4 Contructing caption_iterator for "train" phase Using vanilla Generator initialization: xavier Glove emb of the same size as d_model_caps Traceback (most recent call last): File "./sample/single_video_prediction.py", line 281, in <module> args.device_id, args.prop_generator_model_path, args.pretrained_cap_model_path, args.max_prop_per_vid File "./sample/single_video_prediction.py", line 94, in load_prop_model checkpoint = torch.load(prop_generator_model_path, map_location='cpu') File "/root/anaconda3/envs/bmt/lib/python3.7/site-packages/torch/serialization.py", line 386, in load return _load(f, map_location, pickle_module, **pickle_load_args) File "/root/anaconda3/envs/bmt/lib/python3.7/site-packages/torch/serialization.py", line 580, in _load deserialized_objects[key]._set_from_file(f, offset, f_should_read_directly) RuntimeError: storage has wrong size: expected 0 got 512 I need your help, and I’d appreciate it if you could help me.

[Johnzu-2019]

恐怕我无法重现您的错误。

我建议检查您要加载的模型的md5总和：

md5sum ./sample/best_prop_model.pt
5f8b20826b09eadd41b7a5be662c198b ./sample/best_prop_model.pt

md5sum ./sample/best_cap_model.pt
7b4d48cd77ec49a027a4a1abc6867ee7 ./sample/best_cap_model.pt

您是否正在使用某种形式的分布式培训？

I am afraid I cannot reproduce your error on my side.

I would suggest checking the md5 sums of the models you are trying to load:

md5sum ./sample/best_prop_model.pt
5f8b20826b09eadd41b7a5be662c198b  ./sample/best_prop_model.pt

md5sum ./sample/best_cap_model.pt
7b4d48cd77ec49a027a4a1abc6867ee7  ./sample/best_cap_model.pt

Are you using distributed training of some sort?

Thank you very much indeed. According to your suggestion, I found that the MD5 code of the.pt file I uploaded to the server was wrong. After I re-uploaded the correct.PT file, I ran it successfully.You are such a good man. You have been very helpful to me as a beginner.

[v-iashin]

No problem. Good luck with your endeavors!

Closing this.