Skip to content

CNNT: dynamic positional embeddings to support variable input sizes #150

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SarthakJagota wants to merge 1 commit into
base: main
Choose a base branch
from
Open

CNNT: dynamic positional embeddings to support variable input sizes #150

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 9 additions & 2 deletions DeepLense_Regression_Zhongchao_Guan/CNNT.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -173,10 +173,9 @@ def __init__(self, num_classes, depth, heads, mlp_dim, pool='mean', dim_head=64,
self.to_patch_embedding = CNN()

# hyper-params
num_patches = 16 * 16
dim = 32

self.pos_embedding = nn.Parameter(torch.randn(1, num_patches + 1, dim))
self.pos_embedding = None
self.cls_token = nn.Parameter(torch.randn(1, 1, dim))
self.dropout = nn.Dropout(emb_dropout)

Expand All @@ -196,6 +195,14 @@ def forward(self, img):
# b: batch size n: patch number _: dim of patch
b, n, _ = x.shape

# create pos embedding dynamically
if self.pos_embedding is None or self.pos_embedding.shape[1] != n + 1:
pos_emb = nn.Parameter(
torch.randn(1, n + 1, x.shape[-1], device=x.device)
)
self.register_parameter("pos_embedding", pos_emb)
self.pos_embedding = pos_emb

cls_tokens = repeat(self.cls_token, '1 1 d -> b 1 d', b=b)
x = torch.cat((cls_tokens, x), dim=1)
x += self.pos_embedding[:, :(n + 1)]
Expand Down