About the parameters of train

[znithx]

I want to know the parameter of train you set.
In the coco.yaml file you mentioned that:

model:
name: pcme
embed_dim: 512 # 2048 origin
cnn_type: resnet50 # res152 origin
wemb_type: glove
word_dim: 300

what means of 2048 origin ? I should use 512 or 2048 （resnet50 or res152？） to reproduce your results？

And in the main.py ,,the parameter of feature_dim should be 256 or 512(because you mentioned representation dimension is 512) 。。
parser.add_argument('--batch_size', type=int, default=64, metavar='N',
help='input batch size for training (default: 256)')
the batch_size should be 64 or 256 ?

Looking forward to your reply. Thank you!