search

YunxinLi / Multimodal-Context-Reasoning

A multimodal context reasoning approach that introduce the multi-view semantic alignment information via prefix tuning.
14 stars 0 forks source link

No such file or directory: 'pmr_data/ChunkMaskVal_v4_without_premise.pkl' #5

Open hxan opened 1 year ago

hxan commented 1 year ago

Hello, when I run python run_PMR_ModCR.py --do_train "No such file or directory: 'pmr_data/ chunkmaskval_v4_without_prem.pkl '"Is displayed, and then I replace the above file with ChunkMaskTrain_v4_vcr.pkl in vcr_data and run it again, prompting clip missing. "No such file or directory: 'pmr_data/images/'" is displayed after deleting the clip code.

YunxinLi commented 1 year ago

Hello, In this repositoty, we could only upload some files with small size. chunkmaskval_v4_without_prem.pkl and other pkl files could be downloaded from data_feature.zip: https://huggingface.co/YunxinLi/ModCR_checkpoints/tree/main.

YunxinLi commented 1 year ago

The original images could be dowanloaded from the PMR dataset: https://2030nlp.github.io/PMR/