microsoft / pica Goto Github PK

I run this command：python gpt3_api_okvqa.py --apikey xxx --output_path output --engine davinci --similarity_metric random --n_ensemble 1 --n_shot 16. I don't know why it doesn't occupy the memory of the GPU, and I can't even see the process number.
and Is the whole project just a .py file?

Hello, thank you for doing such an excellent job, how many GPUs did you use for this work?

Hi, thanks for sharing your inspiring work. I met a bug when running your code. The caption tag file train.score.json.tsv can not load. The error message is: json.decoder.JSONDecodeError: Expecting ':' delimiter: line 1 column 466 (char 465)
The other two files for val and test are good. How can I fix the problem with that of the training set?