zengyan-97 / X-VLM

X-VLM: Multi-Grained Vision Language Pre-Training (ICML 2022)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Hello, please ask the train_files in configs/yaml file to "hdfs: // path / to / vg" error, please change how to set up

zhanghehe8 opened this issue · comments

Hello, please ask the train_files in configs/yaml file to "hdfs: // path / to / vg" error, please change how to set up

Please prepare your data and replace it with the path to visual genome dataset. I am not sure if I get your question.

Thank you for your reply, can you provide the right JSON Files?The JSON file in the link cannot be used in RegionTextJsondataSet processing.

Please prepare your data and replace it with the path to visual genome dataset. I am not sure if I get your question.

Thank you for your reply, can you provide the right JSON Files?The JSON file in the link cannot be used in RegionTextJsondataSet processing.

When making region_dataset, since there is no suitable JSON files, the JSON file you provide does not contain Box.

Please read the README file carefully. I did not provide pre-training datasets (it might be available in March.). The JSON file I provided is for fine-tuning only.

I'm glad to receive your reply. I see

Please read the README file carefully. I did not provide pre-training datasets (it might be available in March.). The JSON file I provided is for fine-tuning only.

good job !! Looking forward to your pre-training datasets