Open magicwang1111 opened 1 year ago
I'm curious why the files you trained in the second stage have a size of 6G, and the files I trained are all 45MB,my dataset is 5000 pairs.
This dataset is the data of MiniGPT-4 in the second phase. We didn't make any changes. Thanks.
Sorry, we checked, and our paper also clearly states that it is 5000 pairs, 45M, and it may be incorrectly written on Github. Thank you for your feedback.
Sorry, we checked, and our paper also clearly states that it is 5000 pairs, 45M, and it may be incorrectly written on Github. Thank you for your feedback.
can you supply first pretraining stage model,my gpu is too low cant train.
artgpt has a big boost to photo descriptions in art,I have seen that you create a small (3500 pairs in total) yet high-quality dataset,can you public this dataset?thx