Tensor into CNN

Ezno · November 19, 2020, 2:26pm

I am attempting to do the Kaggle Iceberg Challenge. One approach is to take the image values provided in a JSON and save them to an actual image file. Use that to do a CNN with the normal way in fastai. I am trying to feed the values directly in to save this step (and learn how fastai works a bit better so I can do more custom stuff). Any help would be appreciated.

The general idea is:

Create Tensor with all values + dependent variable
Create dataset
Create DataLoader
Create DataLoaders
Create CNN based on DataLoaders

All steps seem to work, until I go to train the CNN. Below is the dataloaders, CNN, and stack trace. A link to the full notebook for preceeding detail is here: https://github.com/Isaac-Flath/kaggle/blob/main/iceberg_classification/main.ipynb

dls = dl.dataloaders()
def print_batch(batch):
    f, axarr = plt.subplots(3,3)
    cntr = 0
    for row in range(0,3):
        for col in range(0,3):
            axarr[row,col].axis('off')
            axarr[row,col].set_title(f'dep var: {batch[1][cntr]}')            
            axarr[row,col].imshow(batch[0][cntr].view(75,75), cmap='Greys',)
            cntr = cntr + 1
print_batch(dls.one_batch())

So a batch seems to give me what I want, but then when I try to put this into a learner, it doesn’t quite work. I’ve tried changing out the model architecture to a simple one I made, and I still get errors. I can only think that the dataloader isn’t correct, but a batch of the dataloader is just the x,y so I am not sure what else I should be attempting to do there.

learner = cnn_learner(dls,resnet18,loss_func=F.cross_entropy, n_out=1)
learner.fit_one_cycle(1)

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-33-158e79851ed2> in <module>
----> 1 learner.fit_one_cycle(1)

~/anaconda3/lib/python3.8/site-packages/fastai/callback/schedule.py in fit_one_cycle(self, n_epoch, lr_max, div, div_final, pct_start, wd, moms, cbs, reset_opt)
    110     scheds = {'lr': combined_cos(pct_start, lr_max/div, lr_max, lr_max/div_final),
    111               'mom': combined_cos(pct_start, *(self.moms if moms is None else moms))}
--> 112     self.fit(n_epoch, cbs=ParamScheduler(scheds)+L(cbs), reset_opt=reset_opt, wd=wd)
    113 
    114 # Cell

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in fit(self, n_epoch, lr, wd, cbs, reset_opt)
    203             self.opt.set_hypers(lr=self.lr if lr is None else lr)
    204             self.n_epoch = n_epoch
--> 205             self._with_events(self._do_fit, 'fit', CancelFitException, self._end_cleanup)
    206 
    207     def _end_cleanup(self): self.dl,self.xb,self.yb,self.pred,self.loss = None,(None,),(None,),None,None

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _with_events(self, f, event_type, ex, final)
    152 
    153     def _with_events(self, f, event_type, ex, final=noop):
--> 154         try:       self(f'before_{event_type}')       ;f()
    155         except ex: self(f'after_cancel_{event_type}')
    156         finally:   self(f'after_{event_type}')        ;final()

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _do_fit(self)
    194         for epoch in range(self.n_epoch):
    195             self.epoch=epoch
--> 196             self._with_events(self._do_epoch, 'epoch', CancelEpochException)
    197 
    198     def fit(self, n_epoch, lr=None, wd=None, cbs=None, reset_opt=False):

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _with_events(self, f, event_type, ex, final)
    152 
    153     def _with_events(self, f, event_type, ex, final=noop):
--> 154         try:       self(f'before_{event_type}')       ;f()
    155         except ex: self(f'after_cancel_{event_type}')
    156         finally:   self(f'after_{event_type}')        ;final()

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _do_epoch(self)
    188 
    189     def _do_epoch(self):
--> 190         self._do_epoch_train()
    191         self._do_epoch_validate()
    192 

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _do_epoch_train(self)
    180     def _do_epoch_train(self):
    181         self.dl = self.dls.train
--> 182         self._with_events(self.all_batches, 'train', CancelTrainException)
    183 
    184     def _do_epoch_validate(self, ds_idx=1, dl=None):

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _with_events(self, f, event_type, ex, final)
    152 
    153     def _with_events(self, f, event_type, ex, final=noop):
--> 154         try:       self(f'before_{event_type}')       ;f()
    155         except ex: self(f'after_cancel_{event_type}')
    156         finally:   self(f'after_{event_type}')        ;final()

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in all_batches(self)
    158     def all_batches(self):
    159         self.n_iter = len(self.dl)
--> 160         for o in enumerate(self.dl): self.one_batch(*o)
    161 
    162     def _do_one_batch(self):

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in one_batch(self, i, b)
    176         self.iter = i
    177         self._split(b)
--> 178         self._with_events(self._do_one_batch, 'batch', CancelBatchException)
    179 
    180     def _do_epoch_train(self):

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _with_events(self, f, event_type, ex, final)
    152 
    153     def _with_events(self, f, event_type, ex, final=noop):
--> 154         try:       self(f'before_{event_type}')       ;f()
    155         except ex: self(f'after_cancel_{event_type}')
    156         finally:   self(f'after_{event_type}')        ;final()

~/anaconda3/lib/python3.8/site-packages/fastai/learner.py in _do_one_batch(self)
    161 
    162     def _do_one_batch(self):
--> 163         self.pred = self.model(*self.xb)
    164         self('after_pred')
    165         if len(self.yb): self.loss = self.loss_func(self.pred, *self.yb)

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/container.py in forward(self, input)
    115     def forward(self, input):
    116         for module in self:
--> 117             input = module(input)
    118         return input
    119 

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/container.py in forward(self, input)
    115     def forward(self, input):
    116         for module in self:
--> 117             input = module(input)
    118         return input
    119 

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/conv.py in forward(self, input)
    421 
    422     def forward(self, input: Tensor) -> Tensor:
--> 423         return self._conv_forward(input, self.weight)
    424 
    425 class Conv3d(_ConvNd):

~/anaconda3/lib/python3.8/site-packages/torch/nn/modules/conv.py in _conv_forward(self, input, weight)
    417                             weight, self.bias, self.stride,
    418                             _pair(0), self.dilation, self.groups)
--> 419         return F.conv2d(input, weight, self.bias, self.stride,
    420                         self.padding, self.dilation, self.groups)
    421 

RuntimeError: Expected 4-dimensional input for 4-dimensional weight [64, 3, 7, 7], but got 2-dimensional input of size [64, 5625] instead

Ezno · November 20, 2020, 1:58am

Figured it out with a really big chunk of help from Zach,ilovescience, and DrHB in discord.

There were 2 issues, the first was that I was passing my images at 2D arrays (bs * pixels). I needed to pass 4D array (bs,channels,x,y). I reshaped those in my datasets (which flows through to the dataloaders).

x_y = [torch.cat((tensor(img['band_1']),tensor([img['is_iceberg']]))) for img in data]

def get_x(o): return o[:-1].view(1,75,75)
def get_y(o): return tensor(o[-1]).type(torch.float).view(1)

ds = Datasets(x_y,[[get_x], [get_y]],splits=splits)

The second issue was that I was trying to feed a 1 channel image to a model that expected 1 channel. I fixed this with the following code below (copied and pasted from what DrHB sent me)

def alter_learner(learn, n_channels=1):
  "Adjust a `Learner`'s model to accept `1` channel"
  layer = learn.model[0][0]
  layer.in_channels=n_channels
  layer.weight = nn.Parameter(layer.weight[:,1,:,:].unsqueeze(1))
  learn.model[0][0] = layer

learn = cnn_learner(dls,resnet18,loss_func=F.mse_loss, n_out=1)

alter_learner(learn)