Oscar

Oscar copied to clipboard

VQA finetune: test annotation file test-dev2015_qla_mrcnn.json and test2015_qla_mrcnn.json are missing

4

Hi, thank you for releasing the great work! I am working on the VQA task. May I ask where can I find the annotation file `test2015_qla_mrcnn.json` and `test-dev2015_qla_mrcnn.json` to make...

yangapku

Hi, I am trying to reproduce your results for NoCaps challenge. For the VinVL + VIVO model (NoCaps challenge) which object labels (tags) did you use for VIVO pretraining and...

enesmsahin

How to fine-tune retrieval task in another language?

2

Hi, I wonder how I can fine-tune the pretrained model to adapt to tasks (more specifically, retrieval task) in another language like Swedish? What steps do you suggest?

KatieGou

is there the pretrain checkpoint for Oscar+？I only find the checkpoint which has been finetuned on coco_ir

1

imhandsome

I can't train model using 2 GPUs

4

I am trying to train the captioning base model on two Quadro RTX 8000 GPUs, each one with 48GiB RAM. But when I run the command to train the model...

gabrielsantosrv

File 'coco_flickr30k_googlecc_gqa_sbu_oi.lineidx' is Not Found

2

Hi! This file is needed for pretraining on Large corpus, but is not found. Could you share this file? Thanks!

lostnighter

checkpoint of Image Text Retrieval

3

Hello, can you give the checkpoint of Image Text Retrieval, your link in VinVL_MODEL_ZOO. md is wrong, thank you!

QC-LY

VQA object tags are different from image feature

5

Hi, I am currently working on VQA datasets. The VQA fine-tune Oscar-base script from `VinVL_MODEL_ZOO.md` use `--data_label_type mask`, so it will use the text data from `train2014_qla_mrcnn.json` downloaded from https://biglmdiag.blob.core.windows.net/vinvl/datasets/vqa...

kehanlu

When I am trying to run VQA-large fine-tuning, I cannot find the file 'train+val_img_frcnn_feats.pt'. Could you please take a look? Thanks

shizhediao

I have some images that contain a mixture of seen and unseen object classes. I have my own custom object detection model based on YOLOv5, and it is able to...

aliencaocao