Pre-trained model for fine-tuning on OK-VQA

ThalesGroup / ConceptBERT

Implementation of ConceptBert: Concept-Aware Representation for Visual Question Answering

https://github.com/ThalesGroup/ConceptBERT

Other

25 stars 11 forks source link

Pre-trained model for fine-tuning on OK-VQA #11

Open monste12 opened 2 years ago

monste12 commented 2 years ago

Hello,

I'm trying to reproduce this work. Now I'm in the step of fine-tuning on OK-VQA.

How can I get the pre-trained model (pytorch_model_11.bin) required for fine-tuning? /nas-data/vilbert/data2/save_final/VQA_bert_base_6layer_6conect-beta_vilbert_vqa/pytorch_model_11.bin

Thank you.

marlenegautier commented 2 years ago

Hello, The pre-trained model (pytorch_model_11.bin) is now available on the kaggle of the project: https://www.kaggle.com/thalesgroup/conceptbert/

Have a great day

monste12 commented 2 years ago

Thank you so much for uploading the model.

Regarding the result of fine-tuning on OK-VQA, I am wondering why the training score is quite low, around 0.02. Could you please if the result file, ConceptBERT/conceptBert/results/OKVQA_model3-kilbert_training/out.txt, is correct?