wysiwyr
wysiwyr copied to clipboard
yonatanbitton
Training images in the google drive link does not work. Could you fix this?
Hi, thanks for the great work -- any updates on the release of the last two datasets? Pick-a-pic con and COCO-con? Also is there a processing script for SNLI-VE (or...
Hi, Thanks for the excellent work! I have briefly read the code repository and paper, but I was unable find the training data and configuration files for training the End-to-end...
Hi again and thanks for the great work. I noticed that the dataset you released doesn't include coco-con image-text pairs. Is this on purpose? Do you plan to release those?...
Hey, thanks for the great work! I had a question about the training dataset for the end-to-end VNLI model. In the paper you mention: `Specifically, we finetune BLIP2 and PaLI-17B...
Is there a way to replicate the PaLI results? Either a training script, or ideally a checkpoint would be awesome. I'm hoping to use it as a component in a...
Could you please explain how you precisely assess your methods and Blip2 using the Winoground metric?
Hi, I'm trying to replicate the VNLI results. When running ```itm_vqa_blip2_train.py``` it's expecting a model config file for blip2, which isn't included in the repository. Is that available somewhere? ```...
Hi and thanks for the great work :) I'm trying to run the VQ2 pipeline, and can't find the code for calculating the "yes" score in BLIP-2 VQA, the code...
Hi~, the project is so great! However, why is there no picture in the link of the picture in SeeTRUE Dataset?
