Oscar
Oscar copied to clipboard
Oscar and VinVL
We tried to reproduce the baselines for the NLVR2 task. But our result was off by a visible margin. ### Hardware Specifications Graphic Card : Quadro RTX 6000 CUDA version...
I am trying to follow the README file in the Oscar MODEL_ZOO.md and I am encountering a few issues. 1. For finetuning on the tasks of VQA, GQA and NLVR2,...
I downloaded vinvl captioning data from https://github.com/microsoft/Oscar/blob/master/VinVL_DOWNLOAD.md. When I run captioning training: I found the size of features (https://github.com/microsoft/Oscar/blob/master/oscar/run_captioning.py#L138) and the size of label_info(https://github.com/microsoft/Oscar/blob/master/oscar/run_captioning.py#L124) don't match; In principle they should...
Hi, thanks for releasing OSCAR+. I try to download pretrained OSCAR+ model for the task of NoCaps as instructed in [VinVL_DOWNLOAD.md](https://github.com/microsoft/Oscar/blob/master/VinVL_DOWNLOAD.md), using `path/to/azcopy copy 'https://biglmdiag.blob.core.windows.net/vinvl/model_ckpts/nocaps' --resursive` However, the download cannot...
what is the input of nocaps inference? the image or the image feature throngthout vinvl?
I cant found the config file train_yaml="train.fea.penzhan2.lab.oid_X152_min10.yaml" for image captioning, can you provide the download path?
where is the nocaps task trained model?
This is inconsistent with the number reported in the paper
Hello, thanks for releasing such an excellent model. I found that in [VinVL_DOWNLOAD.md](https://github.com/microsoft/Oscar/blob/master/VinVL_DOWNLOAD.md), there is a typro in command `path/to/azcopy copy 'https://biglmdiag.blob.core.windows.net/vinvl/model_ckpts/nocaps' --resursive` . It should be `--recursive` at the...