Maximum value of max_n in show_batch()

moon_knight · March 24, 2022, 2:05am

What is the maximum number of images that can be showed using dataloaders.show_batch()?
Is it 9?

matdmiller · March 24, 2022, 4:43am

I don’t think there is a hard limit. The default is maximum is 9, but that can be overridden by supplying a new max_n value. At some point the figure will get too large so there is probably a practical limit, but I don’t know what it is.

github.com

fastai/fastai/blob/773d5b536c42234a58731a372228ade846804745/fastai/data/core.py#L96

      
        
                return L(batch_to_samples(b, max_n=max_n)).map(f)
            
            
def _pre_show_batch(self, b, max_n=9):
                "Decode `b` to be ready for `show_batch`"
                b = self.decode(b)
                if hasattr(b, 'show'): return b,None,None
                its = self._decode_batch(b, max_n, full=False)
                if not is_listy(b): b,its = [b],L((o,) for o in its)
                return detuplify(b[:self.n_inp]),detuplify(b[self.n_inp:]),its
            
            
def show_batch(self, b=None, max_n=9, ctxs=None, show=True, unique=False, **kwargs):
                if unique:
                    old_get_idxs = self.get_idxs
                    self.get_idxs = lambda: Inf.zeros
                if b is None: b = self.one_batch()
                if not show: return self._pre_show_batch(b, max_n=max_n)
                show_batch(*self._pre_show_batch(b, max_n=max_n), ctxs=ctxs, max_n=max_n, **kwargs)
                if unique: self.get_idxs = old_get_idxs
            
            
def show_results(self, b, out, max_n=9, ctxs=None, show=True, **kwargs):
                x,y,its = self.show_batch(b, max_n=max_n, show=False)

thatgeeman · March 24, 2022, 5:44am

Naturally the upper limit should be the number of images in the batch: b.shape[0]

moon_knight · March 24, 2022, 7:36am

Got it. Thanks for the clarification.

moon_knight · March 24, 2022, 7:38am

Thanks!
Yes, that’s what I was expecting. I read that default batch size is 64 in fast ai so I was expecting the 64 images but I totally forgot that since I am not passing any value for max_n then it’s default value will be taken i.e. 9.