Skip to content
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.

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

Master thesis superRes stuff and RIM #113

Draft
wants to merge 59 commits into
base: main
Choose a base branch
from
Draft
Changes from 1 commit
Commits
Show all changes
59 commits
Select commit Hold shift + click to select a range
c145413
remove splitted_L1
StFroese Nov 23, 2020
4dfcdcc
add psnr and ssim
StFroese Nov 23, 2020
c344996
add EDSR and RDNet
StFroese Nov 23, 2020
12b2faf
Merge remote-tracking branch 'origin/noise_arch' into super_res
StFroese Nov 23, 2020
4e96c46
add Feedback Arch and SRResNet_sym with transformation of input
StFroese Dec 10, 2020
d4e4a06
add lr scheduler as comment
StFroese Dec 10, 2020
2b4f07d
add splitted_l1
StFroese Dec 10, 2020
2bd6dc8
fix for HDF5 lock
StFroese Dec 10, 2020
e646cfe
add Net suffix for define_arch
StFroese Dec 10, 2020
8bbf243
SRResNet_sym_pad and VGG19
StFroese Jan 11, 2021
7c2c15f
SRGAN
StFroese Jan 11, 2021
63f83ae
pytorch prediction
StFroese Jan 11, 2021
d1ca026
SRGAN
StFroese Jan 11, 2021
1246069
SRGAN
StFroese Jan 11, 2021
894dbd6
save SRGAN + new transformation
StFroese Jan 11, 2021
9359440
SRGAN
StFroese Jan 11, 2021
7cbd036
Process VLBI Data
StFroese Jan 11, 2021
dad3edc
env + fine_tune
StFroese Jan 21, 2021
39fdb79
delete gan
StFroese Apr 22, 2021
e81b149
delete gan
StFroese Apr 22, 2021
c5499c5
delete gan
StFroese Apr 22, 2021
8dfda4f
delete gan
StFroese Apr 22, 2021
bfe8dfc
delete gan
StFroese Apr 22, 2021
9dcb8dd
delete gan
StFroese Apr 22, 2021
fce638d
changed for vipy/eht simulations; VERY HARD CODE
StFroese Apr 22, 2021
9cc0009
Merge remote-tracking branch 'origin/main' into super_res
StFroese Apr 22, 2021
61e9450
gridding for model loss
StFroese May 19, 2021
ecd68c9
vgg19 feature loss & training; gan implementation
StFroese May 19, 2021
be58f68
dirty model .h5 file from process_vlbi.py are now readable
StFroese May 20, 2021
8a1daf3
GANCS + CLEANNN
StFroese Jun 14, 2021
43153e0
RNN Callback
StFroese Jun 14, 2021
39786a6
tuple input h5
StFroese Jun 14, 2021
46e231d
GANLearner
StFroese Jun 14, 2021
c2ea72c
GANCS Loss + physics informed loss
StFroese Jun 14, 2021
2982a6d
HardDC + SoftDC + gauss
StFroese Jun 14, 2021
5a338fa
gan save
StFroese Jun 14, 2021
cf597d8
gan save only generator
StFroese Jun 14, 2021
87b85a2
gridding + response + baselines-mask
StFroese Jun 14, 2021
7f05e98
ConvGRUCell
StFroese Jun 14, 2021
4a02a54
ConvGRUCell
StFroese Jun 14, 2021
d7b76c5
ConvGRUCell + gradient Function
StFroese Jun 15, 2021
ca03c07
deactivate OverwriteOneBatch_CLEAN
StFroese Jun 15, 2021
f77c9bb
ConvRNN + RIM
StFroese Jun 15, 2021
5b37b31
RNN Loss function
StFroese Jun 15, 2021
db0d8ca
cleanup gradfunc
StFroese Jun 15, 2021
d2d5e9f
pred change to tuple
StFroese Jun 15, 2021
0da2c3b
RIM
StFroese Jul 15, 2021
8db63fd
rim update
StFroese Nov 30, 2021
a44d71d
loss
StFroese Nov 30, 2021
71a4e8a
minor changes gpu/cpu
StFroese Nov 30, 2021
eac4a42
gridding
StFroese Nov 30, 2021
e53caf2
eht layout
StFroese Nov 30, 2021
bdf49f0
eht layout
StFroese Nov 30, 2021
bf75c6e
markersize antenna distribution changed
StFroese Nov 30, 2021
a393616
merge main
StFroese Nov 30, 2021
77f7425
gridding fix
StFroese Apr 13, 2022
e8ba332
merge from main
StFroese Apr 13, 2022
903672c
delete unused architectures
StFroese Apr 13, 2022
347a7a3
add rim to evaluation.toml
StFroese Apr 13, 2022
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
Prev Previous commit
Next Next commit
cleanup gradfunc
StFroese committed Jun 15, 2021
commit db0d8caf27d8a61e70994d2ca8442a6ccb3e33df
42 changes: 1 addition & 41 deletions radionets/dl_framework/model.py
Original file line number Diff line number Diff line change
@@ -854,44 +854,6 @@ def forward(self, x, hx=None):
return hx

def gradFunc(x, y, A, base_mask, n_tel, base_nums):
# does_require_grad = x.requires_grad
# with torch.enable_grad():
# x.requires_grad_(True)
# print(y.shape)
# base_nums = torch.zeros(base_nums)
# c = 0
# for i in range(n_tel):
# for j in range(n_tel):
# if j<=i:
# continue
# base_nums[c] = 256 * (i + 1) + j + 1
# c += 1
# difference = torch.zeros((x.shape[0],1,x.shape[2],x.shape[3]), dtype=torch.complex64).to('cuda')
# c = 0
# for idx, bn in enumerate(base_nums):
# s_uv = torch.sum((base_mask == bn),3)
# if not (base_mask == bn).any():
# continue

# xA = torch.einsum('bclm,blm->bclm',x,A[...,idx])
# x_prime = xA[:,0] + 1j*xA[:,1] #from 2 channels to complex for fft
# k_prime = torch.fft.fftshift(torch.fft.fft2(torch.fft.fftshift(x_prime)))
# y_prime = torch.einsum('blm,bclm->bclm',s_uv,k_prime.unsqueeze(1))

# # Y = torch.einsum('blm,bclm->bclm', s_uv, y.unsqueeze(1))

# # d = Y - y_prime

# difference += y_prime
# c += 1
# print(difference.shape)
# print(y.shape)
# print((difference-y).shape)
# print((difference-y.unsqueeze(1)).shape)
# points = base_mask.clone()
# points[points != 0] = 1
# points = torch.sum(points,3)
# points[points == 0] = 1
mask = torch.sum(base_mask, 3)
mask[mask != 0] = 1

@@ -905,7 +867,5 @@ def gradFunc(x, y, A, base_mask, n_tel, base_nums):
grad = torch.zeros((ift.size(0), 2) + ift.size()[1:]).to('cuda')
grad[:,0] = ift.real.squeeze(1)
grad[:,1] = ift.imag.squeeze(1)
# grad_x = torch.autograd.grad(spatial, inputs=x, retain_graph=does_require_grad, create_graph=does_require_grad)[0]
# grad_2c = torch.zeros((grad_x, 1) + grad_x()[2:], dtype=torch.complex64).to('cuda')
# x.requires_grad_(does_require_grad)

return grad