AntoineH/BU_Stoch_pool
1
0
Fork 0
mirror of https://github.com/AntoineHX/BU_Stoch_pool.git synced 2025-05-04 09:40:46 +02:00
Code Issues Projects Releases Packages Wiki Activity

added forward fast in stoch.py, but it works only with ceil_mode=False

Browse source
This commit is contained in:
Marco Pedersoli 2020-06-18 12:27:45 -04:00
parent f7436d0002
commit 286921f8a0
3 changed files with 197 additions and 59 deletions
Download patch file Download diff file Expand all files Collapse all files

35
models/mylenet4.py
View file

@ -101,6 +101,30 @@ class MyLeNetMatNormal(nn.Module):#epach 21s
#out = (self.fc1(out)) #out = (self.fc1(out))
return out return out
class MyLeNetMatStochNoceil(nn.Module):#epoch 17s
def __init__(self):
super(MyLeNetMatStochNoceil, self).__init__()
self.conv1 = SConv2dAvg(3, 200, 3, stride=2,padding=1,ceil_mode=False)
self.conv2 = SConv2dAvg(200, 400, 3, stride=2,padding=1,ceil_mode=False)
self.conv3 = SConv2dAvg(400, 800, 3, stride=2,padding=1,ceil_mode=False)
self.conv4 = SConv2dAvg(800, 10, 3, stride=4,padding=1,ceil_mode=False)
#self.fc1 = nn.Linear(800, 10)
def forward(self, x, stoch=True):
#print('in',x.shape)
out = F.relu(self.conv1(x,stoch=stoch))
#print('c1',out.shape)
out = F.relu(self.conv2(out,stoch=stoch))
#print('c2', out.shape)
out = F.relu(self.conv3(out,stoch=stoch))
#print('c3',out.shape)
out = self.conv4(out,stoch=stoch)
#print('c4',out.shape)
out = out.view(out.size(0), -1 )
#out = self.fc1(out)
return out
class MyLeNetMatStoch(nn.Module):#epoch 17s class MyLeNetMatStoch(nn.Module):#epoch 17s
def __init__(self): def __init__(self):
super(MyLeNetMatStoch, self).__init__() super(MyLeNetMatStoch, self).__init__()
@ -111,16 +135,15 @@ class MyLeNetMatStoch(nn.Module):#epoch 17s
#self.fc1 = nn.Linear(800, 10) #self.fc1 = nn.Linear(800, 10)
def forward(self, x, stoch=True): def forward(self, x, stoch=True):
print('in',x.shape) #print('in',x.shape)
out = F.relu(self.conv1(x,stoch=stoch)) out = F.relu(self.conv1(x,stoch=stoch))
print('c1',out.shape) #print('c1',out.shape)
out = F.relu(self.conv2(out,stoch=stoch)) out = F.relu(self.conv2(out,stoch=stoch))
print('c2', out.shape) #print('c2', out.shape)
out = F.relu(self.conv3(out,stoch=stoch)) out = F.relu(self.conv3(out,stoch=stoch))
print('c3',out.shape) #print('c3',out.shape)
#hkjhlg
out = self.conv4(out,stoch=stoch) out = self.conv4(out,stoch=stoch)
print('c4',out.shape) #print('c4',out.shape)
out = out.view(out.size(0), -1 ) out = out.view(out.size(0), -1 )
#out = self.fc1(out) #out = self.fc1(out)
return out return out

2
models/myresnet4.py
View file

@ -26,7 +26,7 @@ class SAvg_Pool2d(nn.Module):
out = savg_pool2d(x, self.stride, mode = self.mode,ceil_mode = self.ceil_mode) out = savg_pool2d(x, self.stride, mode = self.mode,ceil_mode = self.ceil_mode)
return out return out
stochmode = 'sim'#'sim'#'stride''stoch''' stochmode = 'stoch'#'sim'#'stride''stoch'''
finalstochpool = True finalstochpool = True
simmode = 'sbc' simmode = 'sbc'

219
models/stoch.py
View file

@ -45,8 +45,55 @@ class SConv2dAvg(nn.Module):
self.padding = padding self.padding = padding
self.kernel_size = kernel_size self.kernel_size = kernel_size
self.ceil_mode = ceil_mode self.ceil_mode = ceil_mode
def forward_fast(self, input, index=-torch.ones(1), mask=-torch.ones(1,1),stoch=True,stride=-1): #ceil_mode = True not right
device=input.device
if stride==-1:
stride = self.stride #if stride not defined use self.stride
if stoch==False:
stride=1 #test with real average pooling
batch_size, in_channels, in_h, in_w = input.shape
out_channels, in_channels, kh, kw = self.weight.shape
afterconv_h,afterconv_w,out_h,out_w = self.get_size(in_h,in_w,stride)
unfold = torch.nn.Unfold(kernel_size=(kh, kw), dilation=self.dilation, padding=self.padding, stride=1)
inp_unf = unfold(input) #transform into a matrix (batch_size, in_channels*kh*kw,afterconv_h,afterconv_w)
if index[0]==-1 and stride!=1: #or stride!=1:
index,mask = self.sample(in_h,in_w,batch_size,device,mask)
if mask[0,0]==-1:# in case of not given mask use only sampled selection
#inp_unf = inp_unf[:,:,rng_h,rng_w].view(batch_size,in_channels*kh*kw,-1)
inp_unf = torch.gather(inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h*afterconv_w),2,index.view(batch_size,in_channels*kh*kw,out_h*out_w)).view(batch_size,in_channels*kh*kw,out_h*out_w)
else:#in case of a valid mask use selection only on the mask locations
inp_unf = inp_unf[:,:,rng_h[mask>0],rng_w[mask>0]]
#Matrix mul
if self.bias is None:
#flt = self.weight.view(self.weight.size(0), -1).t()
#out_unf = inp_unf.transpose(2,1).matmul(flt).transpose(1, 2)
out_unf = oe.contract('bji,kj->bki',inp_unf,self.weight.view(self.weight.size(0), -1),backend='torch')
#print(((out_unf-out_unf1)**2).mean())
else:
#out_unf = oe.contract('bji,kj,k->bki',inp_unf,self.weight.view(self.weight.size(0), -1),self.bias,backend='torch')#+self.bias.view(1,-1,1)#wrong
out_unf = oe.contract('bji,kj->bki',inp_unf,self.weight.view(self.weight.size(0), -1),backend='torch')+self.bias.view(1,-1,1)#sligthly slower but correct
#out_unf1 = (inp_unf.transpose(1, 2).matmul(self.weight.view(self.weight.size(0), -1).t()) + self.bias).transpose(1, 2)
#print(((out_unf-out_unf1)**2).mean())
#self.flt = self.weight.view(self.weight.size(0), -1).t()
#out_unf = (inp_unf.transpose(1, 2).matmul(self.flt) + self.bias).transpose(1, 2)
if stride==1 or mask[0,0]==-1:# in case of no mask and stride==1
out = out_unf.view(batch_size,out_channels,out_h,out_w) #Fold
if stoch==False: #this is done outside for more clarity
out = F.avg_pool2d(out,self.stride,ceil_mode=True)
else:#in case of mask
out = torch.zeros(batch_size, out_channels,out_h,out_w,device=device)
out[:,:,mask>0] = out_unf
return out
def forward(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1): def forward_test(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1):#ugly but faster
device=input.device device=input.device
if stride==-1: if stride==-1:
stride = self.stride #if stride not defined use self.stride stride = self.stride #if stride not defined use self.stride
@ -55,41 +102,48 @@ class SConv2dAvg(nn.Module):
batch_size, in_channels, in_h, in_w = input.shape batch_size, in_channels, in_h, in_w = input.shape
out_channels, in_channels, kh, kw = self.weight.shape out_channels, in_channels, kh, kw = self.weight.shape
afterconv_h = in_h+2*self.padding-(kh-1) #size after conv #afterconv_h,afterconv_w,out_h,out_w = self.get_size(in_h,in_w)
afterconv_w = in_w+2*self.padding-(kw-1) #if selh[0,0]==-1:
if self.ceil_mode: #ceil_mode = talse default mode for strided conv # index,mask = self.sample(in_h,in_w,batch_size,device,mask)
out_h = math.ceil(afterconv_h/stride)
out_w = math.ceil(afterconv_w/stride) if 1:
else: #ceil_mode = false default mode for pooling afterconv_h = in_h+2*self.padding-(kh-1) #size after conv
out_h = math.floor(afterconv_h/stride) afterconv_w = in_w+2*self.padding-(kw-1)
out_w = math.floor(afterconv_w/stride) if self.ceil_mode: #ceil_mode = talse default mode for strided conv
out_h = math.ceil(afterconv_h/stride)
out_w = math.ceil(afterconv_w/stride)
else: #ceil_mode = false default mode for pooling
out_h = math.floor(afterconv_h/stride)
out_w = math.floor(afterconv_w/stride)
unfold = torch.nn.Unfold(kernel_size=(kh, kw), dilation=self.dilation, padding=self.padding, stride=1) unfold = torch.nn.Unfold(kernel_size=(kh, kw), dilation=self.dilation, padding=self.padding, stride=1)
inp_unf = unfold(input) #transform into a matrix (batch_size, in_channels*kh*kw,afterconv_h,afterconv_w) inp_unf = unfold(input) #transform into a matrix (batch_size, in_channels*kh*kw,afterconv_h,afterconv_w)
if stride!=1: # if stride==1 there is no pooling if 1:
inp_unf = inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h,afterconv_w) if stride!=1: # if stride==1 there is no pooling
if selh[0,0]==-1: # if not given sampled selection inp_unf = inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h,afterconv_w)
#selction of where to sample for each pooling location if selh[0,0]==-1: # if not given sampled selection
sel = torch.randint(stride*stride,(out_h,out_w), device=device) #selction of where to sample for each pooling location
sel = torch.randint(stride*stride,(out_h,out_w), device=device)
if self.ceil_mode: #in case of ceil_mode need to select only the good locations for the last regions if self.ceil_mode: #in case of ceil_mode need to select only the good locations for the last regions
resth = (out_h*stride)-afterconv_h resth = (out_h*stride)-afterconv_h
restw = (out_w*stride)-afterconv_w restw = (out_w*stride)-afterconv_w
if resth!=0: if resth!=0:
sel[-1] = (sel[-1]//stride)%(stride-resth)*stride+(sel[-1]%stride) sel[-1] = (sel[-1]//stride)%(stride-resth)*stride+(sel[-1]%stride)
sel[:,-1] = (sel[:,-1]%stride)%(stride-restw)+sel[:,-1]//stride*stride sel[:,-1] = (sel[:,-1]%stride)%(stride-restw)+sel[:,-1]//stride*stride
#print(stride-resth,sel[-1]) #print(stride-resth,sel[-1])
#print(stride-restw,sel[:,-1]) #print(stride-restw,sel[:,-1])
rng = torch.arange(0,afterconv_h*afterconv_w,stride*stride,device=device).view(out_h,out_w) #rng = torch.arange(0,afterconv_h*afterconv_w,stride*stride,device=device).view(out_h,out_w)
index = sel+rng rng = torch.arange(0,out_h*stride*out_w*stride,stride*stride,device=device).view(out_h,out_w)
index = index.repeat(batch_size,in_channels*kh*kw,1,1) index = sel+rng
index = index.repeat(batch_size,in_channels*kh*kw,1,1)
if mask[0,0]==-1:# in case of not given mask use only sampled selection if mask[0,0]==-1:# in case of not given mask use only sampled selection
#inp_unf = inp_unf[:,:,rng_h,rng_w].view(batch_size,in_channels*kh*kw,-1) #inp_unf = inp_unf[:,:,rng_h,rng_w].view(batch_size,in_channels*kh*kw,-1)
inp_unf = torch.gather(inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h*afterconv_w),2,index.view(batch_size,in_channels*kh*kw,out_h*out_w)).view(batch_size,in_channels*kh*kw,out_h*out_w) inp_unf = torch.gather(inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h*afterconv_w),2,index.view(batch_size,in_channels*kh*kw,out_h*out_w)).view(batch_size,in_channels*kh*kw,out_h*out_w)
else:#in case of a valid mask use selection only on the mask locations else:#in case of a valid mask use selection only on the mask locations
inp_unf = inp_unf[:,:,rng_h[mask>0],rng_w[mask>0]] inp_unf = inp_unf[:,:,rng_h[mask>0],rng_w[mask>0]]
#Matrix mul #Matrix mul
if self.bias is None: if self.bias is None:
@ -105,15 +159,15 @@ class SConv2dAvg(nn.Module):
if stride==1 or mask[0,0]==-1:# in case of no mask and stride==1 if stride==1 or mask[0,0]==-1:# in case of no mask and stride==1
out = out_unf.view(batch_size,out_channels,out_h,out_w) #Fold out = out_unf.view(batch_size,out_channels,out_h,out_w) #Fold
#if stoch==False: #this is done outside for more clarity if stoch==False: #this is done outside for more clarity
# out = F.avg_pool2d(out,self.stride,ceil_mode=True) out = F.avg_pool2d(out,self.stride,ceil_mode=True)
else:#in case of mask else:#in case of mask
out = torch.zeros(batch_size, out_channels,out_h,out_w,device=device) out = torch.zeros(batch_size, out_channels,out_h,out_w,device=device)
out[:,:,mask>0] = out_unf out[:,:,mask>0] = out_unf
return out return out
def forward_(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1): def forward(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1):
device=input.device device=input.device
if stride==-1: if stride==-1:
stride = self.stride #if stride not defined use self.stride stride = self.stride #if stride not defined use self.stride
@ -158,19 +212,19 @@ class SConv2dAvg(nn.Module):
#out_unf = inp_unf.transpose(1, 2).matmul(self.weight.view(self.weight.size(0), -1).t()).transpose(1, 2) #out_unf = inp_unf.transpose(1, 2).matmul(self.weight.view(self.weight.size(0), -1).t()).transpose(1, 2)
out_unf = oe.contract('bji,kj->bki',inp_unf,self.weight.view(self.weight.size(0), -1),backend='torch') out_unf = oe.contract('bji,kj->bki',inp_unf,self.weight.view(self.weight.size(0), -1),backend='torch')
else: else:
dgdg out_unf = oe.contract('bji,kj->bki',inp_unf,self.weight.view(self.weight.size(0), -1),backend='torch')+self.bias.view(1,-1,1)
out_unf = (inp_unf.transpose(1, 2).matmul(self.weight.view(self.weight.size(0), -1).t()) + self.bias).transpose(1, 2) #out_unf = (inp_unf.transpose(1, 2).matmul(self.weight.view(self.weight.size(0), -1).t()) + self.bias).transpose(1, 2)
if stride==1 or mask[0,0]==-1:# in case of no mask and stride==1 if stride==1 or mask[0,0]==-1:# in case of no mask and stride==1
out = out_unf.view(batch_size,out_channels,out_h,out_w) #Fold out = out_unf.view(batch_size,out_channels,out_h,out_w) #Fold
#if stoch==False: #this is done outside for more clarity if stoch==False: #this is done outside for more clarity
# out = F.avg_pool2d(out,self.stride,ceil_mode=True) out = F.avg_pool2d(out,self.stride,ceil_mode=True)
else:#in case of mask else:#in case of mask
out = torch.zeros(batch_size, out_channels,out_h,out_w,device=device) out = torch.zeros(batch_size, out_channels,out_h,out_w,device=device)
out[:,:,mask>0] = out_unf out[:,:,mask>0] = out_unf
return out return out
def forward_(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1): def forward_slowwithbatch(self, input, selh=-torch.ones(1,1), selw=-torch.ones(1,1), mask=-torch.ones(1,1),stoch=True,stride=-1):
device=input.device device=input.device
if stride==-1: if stride==-1:
stride = self.stride stride = self.stride
@ -226,6 +280,7 @@ class SConv2dAvg(nn.Module):
out[:,:,mask>0] = out_unf out[:,:,mask>0] = out_unf
return out return out
def comp(self,h,w,mask=-torch.ones(1,1)): def comp(self,h,w,mask=-torch.ones(1,1)):
out_h = (h-(self.kernel_size))/self.stride out_h = (h-(self.kernel_size))/self.stride
out_w = (w-(self.kernel_size))/self.stride out_w = (w-(self.kernel_size))/self.stride
@ -241,7 +296,7 @@ class SConv2dAvg(nn.Module):
comp = self.weight.numel()*(mask>0).sum() comp = self.weight.numel()*(mask>0).sum()
return comp return comp
def sample(self,h,w,mask): def sample_slow(self,h,w,mask):
''' '''
h, w : forward input shape h, w : forward input shape
mask : mask of output used in computation mask : mask of output used in computation
@ -251,8 +306,8 @@ class SConv2dAvg(nn.Module):
device=mask.device device=mask.device
#Shape after simple forward conv ? #Shape after simple forward conv ?
afterconv_h = h-(kh-1) afterconv_h = h+2*padding-(kh-1)
afterconv_w = w-(kw-1) afterconv_w = w+2*padding-(kw-1)
# print(afterconv_h) # print(afterconv_h)
# print(afterconv_h/stride) # print(afterconv_h/stride)
@ -263,14 +318,9 @@ class SConv2dAvg(nn.Module):
else: else:
out_h = math.floor(afterconv_h/stride) out_h = math.floor(afterconv_h/stride)
out_w = math.floor(afterconv_w/stride) out_w = math.floor(afterconv_w/stride)
# out_h=((afterconv_h+2*self.padding-1)/stride)+1
# out_w=((afterconv_w+2*self.padding-1)/stride)+1
# print('Out',out_h, out_w)
# assert(tuple(mask.shape)==(out_h,out_w))
# out_h,out_w=mask.shape
selh = torch.randint(stride,(out_h,out_w), device=device) #selh = torch.randint(stride,(out_h,out_w), device=device)
selw = torch.randint(stride,(out_h,out_w), device=device) #selw = torch.randint(stride,(out_h,out_w), device=device)
resth = (out_h*stride)-afterconv_h #reste de ceil/floor, 0 ou 1 resth = (out_h*stride)-afterconv_h #reste de ceil/floor, 0 ou 1
restw = (out_w*stride)-afterconv_w restw = (out_w*stride)-afterconv_w
@ -296,8 +346,73 @@ class SConv2dAvg(nn.Module):
fmask = fmask[0,0] fmask = fmask[0,0]
return selh,selw,fmask.long() return selh,selw,fmask.long()
def get_size(self,h,w): def sample(self,in_h,in_w,batch_size,device,mask=-torch.ones(1,1)):
newh=math.floor(((h + 2*self.padding - self.dilation*(self.kernel_size-1) - 1)/self.stride) + 1) '''
neww=math.floor(((w + 2*self.padding - self.dilation*(self.kernel_size-1) - 1)/self.stride) + 1) h, w : forward input shape
return newh, neww mask : mask of output used in computation
'''
stride = self.stride
out_channels, in_channels, kh, kw = self.weight.shape
#device=mask.device
#Shape after simple forward conv ?
afterconv_h = in_h+2*self.padding-(kh-1) #size after conv
afterconv_w = in_w+2*self.padding-(kw-1)
#Shape after forward ? (== mask.shape ?) #Padding, Dilatation pas pris en compte ?
if self.ceil_mode:
out_h = math.ceil(afterconv_h/stride)
out_w = math.ceil(afterconv_w/stride)
else:
out_h = math.floor(afterconv_h/stride)
out_w = math.floor(afterconv_w/stride)
sel = torch.randint(stride*stride,(out_h,out_w), device=device)
if self.ceil_mode: #in case of ceil_mode need to select only the good locations for the last regions
resth = (out_h*stride)-afterconv_h
restw = (out_w*stride)-afterconv_w
if resth!=0:
sel[-1] = (sel[-1]//stride)%(stride-resth)*stride+(sel[-1]%stride)
sel[:,-1] = (sel[:,-1]%stride)%(stride-restw)+sel[:,-1]//stride*stride
rng = torch.arange(0,out_h*stride*out_w*stride,stride*stride,device=device).view(out_h,out_w)
index = sel+rng
index = index.repeat(batch_size,in_channels*kh*kw,1,1)
#inp_unf = torch.gather(inp_unf.view(batch_size,in_channels*kh*kw,afterconv_h*afterconv_w),2,index.view(batch_size,in_channels*kh*kw,out_h*out_w)).view(batch_size,in_channels*kh*kw,out_h*out_w)
if mask[0,0]!=-1:
maskh = (out_h)*stride
maskw = (out_w)*stride
nmask = torch.zeros((maskh,maskw),device=device)
nmask[rng_h,rng_w] = 1
#rmask = mask * nmask
dmask = self.pooldeconv(mask.float().view(1,1,mask.shape[0],mask.shape[1]))
rmask = nmask * dmask
#rmask = rmask[:,:,:out_h,:out_w]
# print('rmask', rmask.shape)
fmask = self.deconv(rmask)
# print('fmask', fmask.shape)
mask = fmask[0,0].long()
return index,mask#.long()
def get_size(self,in_h,in_w,stride=-1):
if stride==-1:
stride = self.stride
out_channels, in_channels, kh, kw = self.weight.shape
afterconv_h = in_h+2*self.padding-(kh-1) #size after conv
afterconv_w = in_w+2*self.padding-(kw-1)
if self.ceil_mode: #ceil_mode = talse default mode for strided conv
out_h = math.ceil(afterconv_h/stride)
out_w = math.ceil(afterconv_w/stride)
else: #ceil_mode = false default mode for pooling
out_h = math.floor(afterconv_h/stride)
out_w = math.floor(afterconv_w/stride)
#newh=math.floor(((h + 2*self.padding - self.dilation*(self.kernel_size-1) - 1)/self.stride) + 1)
#neww=math.floor(((w + 2*self.padding - self.dilation*(self.kernel_size-1) - 1)/self.stride) + 1)
return afterconv_h,afterconv_w,out_h,out_w