make CrossentropySoftmaxArgmax1HotWithBias accept uint* dtype.

theano/tensor/nnet/nnet.py

@@ -752,8 +752,9 @@ class CrossentropySoftmaxArgmax1HotWithBias(gof.Op):
                 or x.type.dtype not in ['float32', 'float64']:
             raise ValueError('b must be 1-d tensor of floats', b.type)
         if y_idx.type.ndim != 1 \
-                or y_idx.type.dtype not in ['int8', 'int16', 'int32', 'int64']:
-            raise ValueError('y_idx must be 1-d tensor of ints', y_idx.type)
+                or y_idx.type.dtype not in ['int8', 'int16', 'int32', 'int64',
+                'uint8', 'uint16', 'uint32', 'uint64']:
+            raise ValueError('y_idx must be 1-d tensor of [u]ints', y_idx.type)
 
 #       TODO: Is this correct? It used to be y, not y_idx
         nll = tensor.TensorType(x.type.dtype,
@@ -887,10 +888,14 @@ class CrossentropySoftmaxArgmax1HotWithBias(gof.Op):
         if ((PyArray_DESCR(%(y_idx)s)->type_num != NPY_INT64)
             && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_INT32)
             && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_INT16)
-            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_INT8))
+            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_INT8)
+            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_UINT64)
+            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_UINT32)
+            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_UINT16)
+            && (PyArray_DESCR(%(y_idx)s)->type_num != NPY_UINT8))
         {
             PyErr_SetString(PyExc_TypeError,
-                 "y_idx not int8, int16, int32, or int64");
+                 "y_idx not [u]int8, [u]int16, [u]int32, or [u]int64");
             %(fail)s;
         }
         if (PyArray_DIMS(%(x)s)[0] != PyArray_DIMS(%(y_idx)s)[0])

theano/tensor/nnet/tests/test_nnet.py

@@ -247,11 +247,15 @@ class T_CrossentropySoftmaxArgmax1HotWithBias(utt.InferShapeTester):
         n_samples = 3
 
         # First test gradient when getting a gradient on the NLL output.
-        def grad_on_nll(x, b):
-            return self.op(x, b, y_idx=numpy.random.randint(
-                low=0, high=n_classes, size=n_samples))[0]
-        utt.verify_grad(grad_on_nll, [numpy.random.rand(n_samples, n_classes),
-            numpy.random.rand(n_classes)])
+        def grad_on_nll_dtype(dtype):
+            def grad_on_nll(x, b):
+                y_idx = numpy.random.randint(low=0, high=n_classes, size=n_samples).astype(dtype)
+                return self.op(x, b, y_idx=y_idx)[0]
+            return grad_on_nll
+        for dtype in ['uint8', 'int8', 'uint64', 'int64']:
+            utt.verify_grad(grad_on_nll_dtype(dtype),
+                            [numpy.random.rand(n_samples, n_classes),
+                             numpy.random.rand(n_classes)])
 
         # Then test gradient when getting a gradient on the softmax output.
         def grad_on_softmax(x, b):