Closed gngdb closed 9 years ago
Should be a simple convnet based on the MNIST example.
Have started working on this now, basically copying the MNIST example and changing it so that we can specify some options. It is here, @matt-graham have any input on this? Will be going over it more carefully soon.
Have made some big changes to this based on the mnist maxout example in the pylearn2 repo. Unfotunately, now getting some new errors. Getting as far as compiling and attempting a single update though, so promising. New error:
/afs/inf.ed.ac.uk/user/s08/s0805516/repos/pylearn2/pylearn2/space/__init__.pyc in _validate_impl(self, is_numeric, batch)
1107 raise ValueError("The value of a IndexSpace batch must be "
1108 "2D, got %d dimensions for %s." % (batch.ndim,
-> 1109 batch))
1110 if batch.shape[1] != self.dim:
1111 raise ValueError("The width of a IndexSpace batch must match "
ValueError: The value of a IndexSpace batch must be 2D, got 1 dimensions for [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 2 2 2 2 2 2 2 2 3 3 3 3 3 4 4 6 6 6 6 6
6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6].
Probably copied with slight modifications from another image recognition problem.