zhanglu

> 为什么训练D的时候不把G的requires_grad设为False呢? > Why not set G's require_grad to False when training D? > > 两次使用了下面的代码，是否可以将下面代码简化？ > Two times using the following code, can you simplify the following code? >...

> I tried to train your complete network under the nvidia GTX 1080Ti with 12G GPU memory. > But pytorch throw out a "out of cuda memory" error. > I...

> ＠zhangluustb. > hi. if i want to train the complete network to reproduce the paper result, what can i do? > Split the network into several parts? Or buy...

> > ＠zhangluustb. > > hi. if i want to train the complete network to reproduce the paper result, what can i do? > > Split the network into several...

目前我采用的标注两次的方式:第一次标类别,第二次将同一关系组的key、value标为同一类别,然后通过转换脚本对两次得到的标注文件进行合并

请问你直接加了一个1*1的卷积的吗？

use this instead: model, _, preprocess = open_clip.create_model_and_transforms('xlm-roberta-large-ViT-H-14', pretrained='frozen_laion5b_s13b_b90k') tokenizer = open_clip.get_tokenizer('xlm-roberta-large-ViT-H-14')

Hi, @djiajunustc . I also cannot access the link of ReferIt splits file. REFERIT_SPLITS_URL="https://s3-sa-east-1.amazonaws.com/query-objseg/referit_splits.tar.bz2" ThanksREFERIT_DATA_URL="http://www.eecs.berkeley.edu/~ronghang/projects/cvpr16_text_obj_retrieval/referitdata.tar.gz" Could you please send me this file? My email address is [[[email protected]](mailto:[email protected])]

@aschleg The v2.0 API says that your current access allows 1,000 requests/day, with a rate limit of 50 requests/second. Is that means it's not possible to use notebook 02 to...